|Publication number||US8010582 B2|
|Application number||US 12/473,070|
|Publication date||Aug 30, 2011|
|Filing date||May 27, 2009|
|Priority date||Mar 18, 2002|
|Also published as||US20030177146, US20090240704|
|Publication number||12473070, 473070, US 8010582 B2, US 8010582B2, US-B2-8010582, US8010582 B2, US8010582B2|
|Inventors||Melvin Richard Zimowski|
|Original Assignee||International Business Machines Corporation|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (26), Non-Patent Citations (2), Referenced by (2), Classifications (5), Legal Events (3)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The present application is a Continuation of U.S. application Ser. No. 10/102,385, filed Mar. 18, 2002. The entire disclosure which is hereby incorporated by reference.
The method, system, and program product described herein are used for migrating large volumes of data from a data source to a data target, and include organizing and inter-relating data or files, e.g., providing an association between different schema and metadata, with, for example, data structure conversion, for optimization of database and file migration and storage, and for data compatibility between different or multiple databases.
Current approaches for migrating large quantities (for example, gigabytes) of data from relational data stores on various database platform sources, such as Windows NT, AIX, Linux, Solaris, and other database platforms, to databases on disparate target platforms, such as IBM “DB2 UDB for OS/390,” do not perform well and are overly complex. Poor performance and complexity can be attributed to a variety of causes including (1) using SQL in a multi-step process to both extract the data from the source data store and place the data into the target data store, (2) storing the data in one or more intermediate data stores (e.g. local file systems) during data migration, (3) performing multiple datatype conversions as the data is moved between environments, (4) using less efficient and inefficient data transfer mechanisms, and (5) using data migration processes that are not fully automated.
The method, system, and program product described herein use bulk data loading interfaces to migrate the source data into a target data store. The method, system, and program product of the invention eliminate the use of intermediate data stores on the target system, execute the extraction of the data from the source data store and the placement of the data into the target data store in parallel, minimize the datatype conversions that are performed, transfer the data in large blocks, and perform the migration in a fully automated fashion. This approach reduces the complexity and the elapsed time required for performing these types of data migrations.
Specifically, the method, system, and program product migrates data from a relational table in a data source to a relational table in a data target. This is accomplished through the use of two parallel batch processes. One of these batch processes unloads the subject data from the data source and uploads the subject data to the data target. The other batch process loads the subject data into the data target. The “unloading” batch process operating on the data source may use an SQL “SELECT * FROM tablename” command to identify the data to be unloaded, where “tablename” is the name of the source relational table. The “loading” batch process operating on the data target may use a LOAD Utility or other bulk data loading mechanism to load the source data into the data target.
The unload of a particular database table at the source and the subsequent load of the database table at the target execute in parallel. The method, system, and program product allow for the parallel unload and load of the database table by eliminating the use of intermediate data files between the two batch processes. Data flows directly from the unload batch process to the load batch process, thereby reducing both the elapsed time and the complexity of the data migration. The method, system, and program product of the invention also permit a write of the data being migrated to disk on the target system, in parallel with the load of the database table. This copy allows the load of the data to be restartable from the last internal load utility checkpoint in the event of a load utility failure.
The method, system, and program product may be used to facilitate migration of disparate source database data, such as Oracle, Sybase, or Microsoft SQL Server, to, for example, IBM “DB2 for OS/390.” In some situations data type conversion may be required. For example, Oracle long data may need to be mapped to IBM UDB DB2 varchar data.
Overall, according to the method, system, and program product of the invention, the complexity of generating SQL view mapping structures, generating batch jobs and coordinating and executing the migration task is reduced due to automated processes and tools.
The method, system, and program product of the invention are intended for use in both peer to peer and client server environments.
The program product of the invention is in the form of computer code on suitable media (such as magnetic or optical media that is machine readable for transmission, installation, instantiation, or execution). The code may be in the form of executable code, in the form of code that requires loading to memory for execution, or (physical or electronic) distribution code that requires one or both of decompression and decryption to be executable.
Aspects of the invention are illustrated in the FIGURES.
The method, system, and program product described herein provides for the migration of relational tables from a data source to a data target. This is accomplished through the unloading of a particular database table at the data source and subsequent loading of the database table at a data target. Through the use of a data pipe between the parallel batch processes for (1) unloading the source data and (2) loading the target data, the data being migrated never needs an intermediate write to disk on the target, thereby reducing both the elapsed time and the complexity of the data migration.
The two steps, 21 and 23, in
The target system, 3, is associated with a database management system, 43, preferably a relational database management system, an associated database, 45, preferably a relational database, application code to unload and upload data from the data source, and a utility program to load data into the source target.
As illustrated in
DMTFUNLD identifies the data to be unloaded through the execution of an SQL statement. Through the use of DRDA, the SQL statement can be executed against a remote DBMS source, and through the use of DB2 UDB Relational Connect, the remote DBMS source can be one of a variety of DBMS products. The SQL statement issued to unload data from the source DBMS is of the form SELECT * FROM tablename. Once DMTFUNLD issues the SQL statement, the data is uploaded to the target system using DRDA blocking.
DMTFUNLD drives the unload and load processes and executes as a batch program.
DMTFUNLD has the psuedocode representation:
Open the SYSIN dataset and the SYSPUNCH pipe
Read table name from SYSIN dataset
Build the SQL statement to be executed: SELECT * FROM
Build the LOAD Utility control statement for the load of the data and
write it to SYSPUNCH
Close the SYSPUNCH pipe
Open the SYSREC pipe
Execute the SELECT statement to retrieve and upload the data to be
migrated and write the data to SYSREC
Close the SYSREC pipe
Close the SYSIN dataset
As illustrated in
The method, system, and program product upload the selected data to the data target using Distributed Relational Database Architecture (DRDA) flows, and then generate the LOAD control statement for the load of the extracted table from the source system, 1, into the target system, 3. The load control statement and the unloaded data are passed from DMTFUNLD, 37, to the DB2 LOAD utility, 41, using batch pipes, 39.
To be noted is that the first, 21, and second, 23, batch jobs execute in parallel. The use of batch pipes, 39, permits (i) upload of data from the source system, 1, and (ii) loading of data into the target system, 3, to be substantially simultaneous. The use of batch pipes, 39, also reduces I/O operations and the use of Direct Access Storage Devices (DASD) by transferring data through processor storage rather then through DASD.
Two batch jobs are required for each table to be migrated. The first batch job, 21, performs the unload of the data from one or more tables of the source, 1, and uploads the extracted data to the DB2 for OS/390 server, 43, on the target, 3. For each table, a second batch job performs the load of the data into a DB2 for OS/390 table using the LOAD Utility, 41. The second batch job for each table, 23, is automatically submitted for execution by the first batch job, 21. The first, 21, and second, 23, batch jobs for an individual table execute in parallel. In addition, multiple first batch jobs can execute in parallel. The methodology also includes a process for generating the Job Control Language (JCL) for the batch jobs.
The unload environment on the source data platform, 1, requires special setup and configuration. The system, method, and program product include a process for accomplishing this prior to the execution of the batch jobs that migrate the tables.
The data being migrated always resides in memory and never needs to be written to an intermediate file or any other type of intermediate data store on the target machine, e.g., an IBM OS/390 machine. Source to target datatype conversions, for example, Oracle to DB2 datatype conversions, occur once, at the time the data is unloaded from the source database. The DRDA protocol is used to transfer the data.
The method, system, and program product of the invention also has the capability to create a copy of the data being loaded—in parallel with the execution of the extraction from the source and the load to the target. This copy is useful in situations where the amount of data being migrated is large and the LOAD Utility, 41, for whatever reason, fails to execute to completion, as the copy can be used to restart the load of the data from the last internal LOAD Utility checkpoint rather than from the start of the load. When used, this optional facility does create a copy of the data in an intermediate data store, the local target file system, for example, an IBM OS/390 file system.
The data migration process is initiated by the batch application (DMTFUNLD), 37, that executes under the target's operating system, for example IBM z/OS. Each invocation of the application:
The LOAD Utility control statement has the form:
The batch application passes the LOAD Utility control statement and the data to the target's database management system's LOAD utility using Batch Pipes, 39. The use of Batch Pipes:
The bulk data migration method, system, and program product makes the following assumptions about the source and target schemas:
These assumptions permit the existing middleware applications to access the tables that have been migrated to the target system, for example, DB2 for OS/390 tables, and preserve the referential integrity relationships that are introduced and managed by those middleware applications.
This example illustrates the creation, modification, and execution of various commands, jobs, and SQL statements used to establish and configure the execution environment for the data migration tool.
The execution environment had the following characteristics:
The data migration tool used the federated database support provided by DB2 UDB V7 and the federated database support for Oracle data sources provided by DB2 Relational Connect. DB2 UDB and Relational Connect were configured for access to Oracle data. The steps included installing Relational Connect and adding an Oracle data source to a federated system, as described in “Setting Up a Federated System to Access Oracle Data Sources” in the chapter “Configuring a Federated Database System” of the DB2 UDB and DB2 Connect Installation and Configuration Supplement.
In order to add Oracle data sources to the federated system, the following steps were performed:
Note: Due to schema restrictions, for the tables migrated in this example, Oracle Long data types had to be cast to DB2 UDB for z/OS varchar(length) data types, where length was the max actual length of the data contained in the Oracle Long column.
Next, it was necessary to modify the configuration for the DB2 UDB Database.
To update the database configuration for the target database, we used the DB2 Command Line Processor to set the default application heap parameter (APPLHEAPSZ) for this database to 1024:
Configuring DB2 for OS/390
This step included inserting a row into the SYSIBM.USERNAMES table with a column TYPE value of “O” for outbound translation, an AUTHID value of blank indicating that translation applies to all authorization ids, a LINKNAME value identical to the SYSIBM.LOCATIONS LINKNAME column value, and a userid and password values respectively for the NEWAUTHID and PASSWORD columns that specify a user ID and password of a user that has the authority to access the Oracle tables via the nicknames and views created above.
While the invention has been described and illustrated with respect to certain preferred embodiments and exemplifications, it is not intended to limit the scope of the invention thereby, but solely by the claims appended hereto.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5678044||Jun 2, 1995||Oct 14, 1997||Electronic Data Systems Corporation||System and method for improved rehosting of software systems|
|US5761667||Aug 7, 1996||Jun 2, 1998||Bmc Software, Inc.||Method of optimizing database organization using sequential unload/load operations|
|US5890159||Nov 10, 1997||Mar 30, 1999||International Computers Limited||Data transfer mechanism between databases using a separate pipe at each database|
|US5899990||Mar 31, 1997||May 4, 1999||Sun Microsystems, Inc.||Java-to-Database Connectivity Server|
|US5937415||Dec 11, 1996||Aug 10, 1999||Sybase, Inc.||Data base development system with methods facilitating copying of data from one data source to another|
|US6016501||Mar 30, 1998||Jan 18, 2000||Bmc Software||Enterprise data movement system and method which performs data load and changed data propagation operations|
|US6029178||Mar 31, 1998||Feb 22, 2000||Bmc Software||Enterprise data movement system and method which maintains and compares edition levels for consistency of replicated data|
|US6035307||Mar 30, 1998||Mar 7, 2000||Bmc Software||Enterprise data movement system and method including opportunistic performance of utilities and data move operations for improved efficiency|
|US6092086||Mar 31, 1998||Jul 18, 2000||Bmc Software||System and method for handling backout processing during capture of changed data in an enterprise computer system|
|US6115704||May 9, 1994||Sep 5, 2000||Bmc Software, Inc.||Extended SQL change definition language for a computer database system|
|US6115722||Dec 9, 1998||Sep 5, 2000||Computer Associates Think, Inc.||Method for checking tablespaces involved in referential integrity|
|US6151602||Mar 25, 1998||Nov 21, 2000||Inprise Corporation||Database system with methods providing a platform-independent self-describing data packet for transmitting information|
|US6151608 *||Apr 7, 1998||Nov 21, 2000||Crystallize, Inc.||Method and system for migrating data|
|US6167405||Apr 27, 1998||Dec 26, 2000||Bull Hn Information Systems Inc.||Method and apparatus for automatically populating a data warehouse system|
|US6170045||Apr 30, 1997||Jan 2, 2001||International Business Machines Corporation||Cross-system data piping using an external shared memory|
|US6219660||Sep 8, 1999||Apr 17, 2001||International Business Machines Corporation||Access path selection for SQL with variables in a RDBMS|
|US6219662||Jul 9, 1998||Apr 17, 2001||International Business Machines Corporation||Supporting database indexes based on a generalized B-tree index|
|US6490590||Feb 14, 2000||Dec 3, 2002||Ncr Corporation||Method of generating a logical data model, physical data model, extraction routines and load routines|
|US6567823||Aug 7, 2000||May 20, 2003||Corigin Ltd.||Change propagation method using DBMS log files|
|US6643671||Aug 27, 2001||Nov 4, 2003||Storage Technology Corporation||System and method for synchronizing a data copy using an accumulation remote copy trio consistency group|
|US6789096||Jun 25, 2001||Sep 7, 2004||Informatica Corporation||Real time sessions in an analytic application|
|US7251747 *||Sep 19, 2002||Jul 31, 2007||Ncr Corp.||Method and system for transferring data using a volatile data transfer mechanism such as a pipe|
|US20010013038||Mar 16, 2001||Aug 9, 2001||Purcell Keith J.||Technique for providing a universal query for multiple different databases|
|US20020161748||Jul 24, 2001||Oct 31, 2002||International Business Machines Corporation||Data loading from a remote data source|
|JP2000029891A||Title not available|
|JP2000029898A||Title not available|
|1||*||"Datajoiner: A Multidatabase Server Version 1", IBM Corporation, White Paper, May 1995, pp. 1-18.|
|2||Jang et al., "An Extended Model for Intergration between the Oracle DBMS and WWW", International Conference on Information and Communications and Signal Processing, ICICS '97, Singapore, Sep. 9-12, 1997, pp. 569-572.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US8335722||Sep 26, 2011||Dec 18, 2012||Google Inc.||Local shopping and inventory|
|US8566137 *||Mar 22, 2011||Oct 22, 2013||Amazon Technologies, Inc.||Inventory across multiple marketplaces|
|U.S. Classification||707/809, 707/803|
|Apr 10, 2015||REMI||Maintenance fee reminder mailed|
|Aug 30, 2015||LAPS||Lapse for failure to pay maintenance fees|
|Oct 20, 2015||FP||Expired due to failure to pay maintenance fee|
Effective date: 20150830